THE IMPORTANCE O F ACCURATE SAMPLING I N THE PRO-

DUCTION AND USE OF CERAMIC MATERIALS*

B Y 1'. 1'. K S K I I T , J K .
AI3STKACT
A discussion of thc theory and methods of sampling and thrir application t o the
production and iise of feldspar and other ceramic matcrials is presented. Analytical
data conccrning segregation within shipments and bin contents arc given.

1. Introduction
Sampling is the process of selecting a representative portion of a lot
of material. An accurate sample is one in which all of the components
are present in the same proportions and states as they occur in the original
lot.
The importance of correct sampling is too often disregarded and too
often carelessly done. I t is much easier to take a handful from a bin of
ground feldspar than it is to take portioris systematically from all sections
of the bin.
11. Theory a n d Methods
T h e principal elements in sampling are the weight of the sample and
the method of taking and handling it.
T o sample a given lot of material properly, i t is
(1) Weight of Sample
necessary to know how small a portion m a y be
taken and still keep within the allowable limits of error.
In an accurate sample a certain number of pieces must be present re-
gardless of the size of those pieces, b u t depending upon the complexity of
the material t o be sampled. Obviously when all pieces are identical in
composition, one is enough for a sample, b u t where the components differ
greatly many pieces are necessary.
Obviously many (i-inch pieces weigh far more than the same number of
one inch, and therefore different weights of samples are necessary for
different sizes of material.
Tables have been established giving weights of samples necessary to
attain accuracy with given sizes and complexities of material. 'I'o provide
for uncertainties, the weights used should be those known to be accurate
for even greater complexity than the particular lot in question. Since
the pieces are always of unequal sizes the tables of weights of the sample
are based on the size of the largest pieces in the lot.
Table I gives minimum sample weights for feldspar and quartz, re-
spectively, the most and least complicated o f ceramic materials.
An accurate sample of G inch and smaller feldspar pieces weighs a t least
4632 pounds. To obtain a 129-pound sample the whole 2 l I 2 tons must be
crushed t o 1 inch and finer.
.~
* Presented at the Annual ?rleeting, AMERICAS CEKAMICSOCIETY, Washington
D. C., February, 1932 (White Wares Division).
444
 ACCURATE SAMPLING 445

To get a 10-pound sample the 129 pounds must be crushed t o 1/4 inch,
and to obtain a 1/2-gram sample the 10 pounds must be reduced to 200-
mesh.
A sample of correct weight for its size of particles is
(2) Segregation
not necessarily an accurate sample. It is inaccurate
unless the method of taking it insures its being representative of the lot in
question. Segregation must be carefully taken into account.

TABLE I
MINIMUM
SAMPLE
WEIGHTS
AT DIFFERENT
PARTICLE
SIZES
Diameter of Pounds
lilrgest piece A B
Inches Mesh Feldspar Quartz
12 18,528 4819
6 4.632 1177.2
4 2;064 523.2
2 516 130.8
1 129 32.7
0.75 72.4 18.4
0.5 32.2 8.1
0.25 8.05 2.04
0.131 6 2.21
0.065 12 0.55
0.0328 20 0.14 0.035
0.0164 40 0.034 0.0088
0,0082 70 0.0087 0.0022
0.0058 100 0.0043 0.0011
0.0041 150 0.0022
0.0029 200 0.0011
NOTE: Weights at different sizes are taken, proportional to the squares of the di-
ameters, starting in each case with '/2-gram samples, at 200- and 100-mesh, respectively.
The feldspar table has been found reliable in practical operation. The allowable error
is within that of a high standard of accuracy for silicate analysis.

A solution is homogeneous and, for practical purposes, clay in a state

of fine suspension may be considered free from segregation, so that a small
portion taken at random would be an accurate sample. But when dry,
all ceramic materials vary, due t o segregation of particle sizes and usually
of particle qualities. For this reason a random scoopful from a car of
clay is not an accurate sample although i t is of sufficient weight. Sampling
methods must always be such as t o take care of segregation accurately.

111. Preliminary Sampling

The most common of hand-sampling methods is
(1) Grab Sampling
taking grab samples. Small and supposedly equal
portions are taken with scoop, shovel, or hand at regular or random
interval: from the lot of material. It is cheap, quick, but not accurate.
In grab sampling of bulk carloads of ceramic materials, grabs shouId be
taken from 3 parallel rows and at least 12 places in all, or from one row
which zigzags 3 times across the car from corner to corner. Further-
446 KNIGHT

more, such samples should alternately be from the top, middle, and
bottom of the car, and the portions should be equal in amount.
Grab samples of coarse, unground materials are unreliable.
Pipe sampling is satisfactory for sampling fine ma-
(2) Pipe Sampling
terial in cars and bins. Sample - points
- should be
evenly distributed similar t o the plans for grab sampling. A pipe sampler
has the advantage of taking equal columns of the material.
NOTE: The simplest form of sampler consists of a piece of pipe with the lower end
sharpencd and the upper end fitted with handles. It should be long enough to reach the
bottom of the car or bin. Several modifications are in use, one being slotted with one
edge of the slot turned out to force the material inside whcn the pipe is inserted and
turned. Another has two slotted pipes, one within the other, the slots coinciding when
the pipe is inserted. The outcr pipe is then turned to seal the inner slot, and with-
drawn. Still another type is larger a t one end than the other to allow the material to
enter more readily. Small pipe samplers are used to sample material in sacks or barrels.
In cases where the material is tightly packed, as with ball clay, an auger sampler similar
to an ordinary earth auger is recommended.

Fractional shoveling may be used in sampling

(3) Fractional Shoveling
while loading or unloading material. One
out of every so many shovelsful is thrown aside for the sample. I t is
not recommended where there are pieces over 2 inches in diameter. The
number of shovelfuls to be taken depends upon conditions. Every 100th
shovelful should be adequate for shipments of fine material. The re-
sulting sample could then be reduced by coning and quartering t o the
proper size for the laboratory.
Hand-sample cutters can be used where ma-
(4) Hand-Sample Cutters
terial flows through spouts if they are
operated to conform t o the rules for accurate manipulation of machine-
sample cutters and are of similar design. Hand cutters require frequent
attention and introduce the difficultly controlled human element.
The coarse material must be crushed, thoroughly
(5) Reduction of
mixed, and coned and quartered, to yield a weight of
Initial Sample
material which is adequate for the resulting maximum
particle size.
Fine material should be mixed, coned, and quartered, or put through a
Jones splitter enough times to leave a small laboratory sample.
Coning and quartering is seldom correctly
(6) Coning and Quartering
done. It may be used, not without great
labor, on lots up t o 50 tons.
In coning, after crushing t o finer than 2-inch pieces, the material is
shoveled into a conical pile, each shovelful being carefully dropped di-
rectly on the cone apex t o obtain symmetrical distribution about the axis.
Equal shovelfuls should be taken from successively different portions
of the initial pile and dumped on the apex from successively adjacent
 ACCURATE SAMPLING 447

positions around the cone. The fines are then swept up and placed on
top of the cone.
Starting a t points near the bottom of the finished cone the material
is dragged radially downward to form a circular, flat cake, working around
the cone. The cake is marked off into equal quarters and the two opposite
quarters are rejected. The remaining quarters are then shoveled into a
pile and the whole process is repeated.
At best, coning and quartering is subject t o error, and for samples up
to one or two hundred pounds, a Jones splitter is more accurate and more
simple.
Machine sampling obviates the personal-
(7) Machine Sampling
element errors, shortens the time required, and
reduces the labor. It is limited in application because the material must
be in motion and must be presented to the samplers in a stream.
Machine samplers are of two types: those which take part of the
stream all of the time and those which take all of the stream a t regular
intervals.
The first type is unsatisfactory, and not all of the second type conform
to accuracy requirements. Some machine samplers under certain con-
ditions are less accurate than good hand methods.
The approved machine has a scoop which moves a t a uniform speed
across and entirely out of the stream of material. The distance between
cutting edges must permit ready access of the coarsest pieces and the
scoop must be deep enough to prevent particles from bouncing out. The
stream should be constant. The cutting intervals should be frequent
enough to catch changes in the material of the stream.
In sampling 80-tori lots of crude feldspar
(8) Sampling Crude Feldspar
the mine product containing large pieces
is crushed to - 3 / 4 inch. A 21/2yo-rnachine sample cut is roll-crushed to
- 1/4 inch and then a 21/2yo-machinesample, or 100 pounds, is put through
a Jones splitter to obtain a small laboratory sample.

IV. Final Sampling

The laboratory sample is again crushed if need be, and further divided
by riffles or by coning and quartering.
The riffles consist of a series of parallel troughs with open spaces
between them. The sample is spread over the troughs with a flat shovel
or pan, care being taken not t o heap i t above the troughs. Either the
portion falling between the troughs or that which falls in them may be
rejected.
The Jones splitter has no open spaces between the troughs, but the
bottoms are steeply inclined first in one direction, then the other, so that
448 KNIGHT

the material is divided and goes to two underneath pans. The riffle
causes less loss from dust than does the Jones splitter.
V. Sampling in the Production of Ceramic Materials
The rapid advancement in feldspar control is due to accurate sampling.
The uniformity and quality of feldspar shipments are higher than ever
before, especially the uniformity. The feldspar mines of &day, however,
are no larger and no more uniform than formerly. Eye analysis of a pile
of crude feldspar is no more accurate today than it was ten years ago.
The sole reason for increased uniformity and quality is accurate sampling
of the crude feldspar followed by accurate tests and analyses. Four
feldspar mills now have automatic sampling units. Some of the other
mills have improved in uniformity by the guidance of analyses of their
ground product but not to the extent possible with machine sampling
before grinding.

VI. Sampling in the Use of Ceramic Materials

Shipments of raw materials are not homogeneous lots of a definite
material. All portions of a car of kaolin are not the same. A shipment of
crude feldspar will vary appreciably from one end to the other and segrega-
tions will occur in a shipment of ground feldspar depending upon its kind,
the number of bins used in processing, and the method of grinding. Feld-
spars high in free quartz and with a potash-to-soda ratio of less than 3 to 1
are more apt to show such a variation. Hoppered bins are excellent
mixers and materially assist in reducing segregation. Continuous grinding
gives better mixing and less segregation of lots of over 2 tons than does
batch grinding.
Although variations are likely to be small in the case of fine materials,
it is best for the consumer to use good sampling methods.
If to avoid segregation of quality is not sufficient
(1) Segregation of
reason for the consumer’s use of accurate sampling
Particle Sizes
methods, the inevitable segregation of particle sizes
in all raw materials is alone reason enough to condemn the all too common
practice of taking a grab sample from the surface of the carload or bin of
material. Coarse particles always tend to settle to the bottom in any
lot of material.
It should further be kept in mind that the
(2) Relation of Com-
coarse portion is quite liable to differ from
position to Particle Size
the fine portion in kind, analysis, and be-
havior of particles. This is true in clays, ground feldspar, and other
ceramic materials.
Table I1 shows the differences between coarse and fine portions of a
lot of 200-mesh commercial feldspar. A is the analysis of the original
 ACCURATE SAMPLING 449

sample. B is the analysis of the particles over 2 0 p in diameter which

were separated by elutriation. D is the analysis of the particles below
2p which were siphoned, while in suspension, from a settling tube. Analy-
sis C of the portion between 20 and 2 p in size was calculated from A, B,
and D. The mineral compositions were calculated from the analyses.
TABLEI1
ANALYSES
A B c
(70) i%) (Y")
SiO, (i9 30 fig. 80(is.78 ( X i . 65
A1203 17.30 17.12 17.46 18.92
FeZOS 0.10 0 . 08 0.13 O.,O(i
CaO 0.42 0.40 0.44 0.60
MgO Tr. T r. T r. T r.
KaO 9 68 9.72 9 ti0 10 20
Na20 3.18 2.98 :<.12 3.47
Loss 0.21 0.18 0.25 0.25
~ ~ _ _ _ ~ ~ . ~~ ___
l00.1FI 100.28 100. 07 1 0 0 15
MINERAL COMPOSITIONS
Microcline 57.4 5 7 .6 5fi. 9 (i0. f i
Albite 27.0 25.5 29.0 29.5
Anorthite 2.1 2.1 2.2 3 .0
Quartz 11.9 13.3 10.1 4.(i
Kaolin, etc. 1.6 1.6 1.8 2.3
Analyses of coarse and fine portions of 200-mesh commercial feldspar.
A = original material
B = portion with particles coarser than 20p diam. 55.00%
C = portion with particles between 20p and 2 p diam. 43.03%
D = portion with particles finer than 2 p diam. 1.97%

The major difference between the coarse and fine portions of Table I1
is in their respective quartz content. Quartz, although friable, strongly
resists further fine grinding after reduction to 0.04 mm. diameter. Feld-
spar particles are more easily ground t o finer sizes. This difference in
hardness accounts for the concentration of feldspar in the extreme fines,
and the concentration of quartz in the coarser portion.
It is also interesting to note from Table I1 the tendency toward con-
centration of soda feldspar in the extreme fines.
NOTE: Since unaltered soda feldspar is generally tougher than potash feldspar, this
concentration of soda feldspar in the extreme fines is harder to explain and is not to be
generally applicable until confirmed by tests of other feldspars.

More highly altered feldspars than those in Table I1 would in all prob-
ability show a much greater concentration of kaolin in the extreme fines.
Another interesting point in the analyses is the concentration of iron
in C. The reason is probably much the same as in the case of the quartz,
garnet in particular having similar grinding properties. The original
sample contained very little mica, but were more of it present it would
be most largely in the coarse portion.
450 KNIGHT

Table I1 in connection with segregation of

(3) Importance of Particle
particle sizes adds emphasis to the need of
Size in Ceramics
accurate sampling. It also has a distinct
application to research work. The importance of particle size in ceramic
materials deserves greater attention. It is quite likely that lack of such
attention accounts in some part for the conflict existing between the re-
ported results of so many ceramic investigations. Research data will
become vastly more valuable when they include accurate chemical,
microscopic, and particle-size analyses of the materials involved. To
make such analyses accurate, sampling is a prime requisite.

VII. Sampling Methods for the User

The user generally receives ceramic materials in a powdered state.
Methods of sampling them as given by the Standards Committee of the
AMERICANCERAMIC SOCIETY are inconsistent. For whiting, the Com-
mittee recommended that cores be taken from 25% of the containers
with a sampling tube, and for lime that 3% of the containers be dumped
and sampled by fractional shoveling, 1 shovel for 3 tons.
It is difficult enough to have men sample accurately without having
several methods for them to follow. It is best that one process be followed
with all shipments of fine materials, basing the process on that adapted
for accurate sampling of the most variable material. Because of the
difficulty of getting men to carry out fractional shoveling and coning and
quartering properly, it is recommended that pipe samplers be used for
shipments both in containers and in bulk.
At least 5% of the containers should be so sampled. Bulk carloads
should be sampled at 12 places in a 30-ton car and 15 in a 50-ton car,
arranged in 3 parallel rows and equidistant in each row. The pipe should
be inserted to the bottom to secure a representative column.
For ball clay an auger is recommended.
All initial samples should be mixed and preferably reduced to the proper
size with a Jones splitter.
Proper labelling is to be insisted upon. Sampling should be done or
supervised by the laboratory.
VIII. Conclusion
Accurate sampling requires conscientious application of a rigid and
correct routine because sampling errors are not easily detected. Segrega-
tion of particle sizes is sufficient reason alone for accurate sampling, espe-
cially since the coarse and fine portions of ceramic materials are found to
differ in composition. This fact is added argument for the growing
belief that particle size of ceramic materials deserves much more atten-
Jour. Amer. Ceram. Soc., 11 [6], 442 (1928).
 TUNNEL KILN CAR LUBRICATION 451

tion. Accurate sampling is the first step in control of production. The

expenditure of time and money in testing inaccurate samples is largely
wasted, and plant control suffers in consequence.

Bibliography
EDITOR’S NOTE: There are scores of references that could be cited but only the
following are given for the benefit of our members not in touch with technical libraries.
1. A. F. Taggart. Handbook of Ore Dressing, Sect. 21. John Wiley & Sons, Inc.
2. G. S. Pope, “Sampling Delivered Coal,” U. S. Bur. Mines, Bull., No. 116.
3. C. R. Forbes, “Grading and Sampling of Diaspore Clays,” Jour. Amer. Cerum.
Soc., 14 [ 5 ] ,382 (1931).
4. Robert Twells, Jr., “Handling and Storing Raw Materials to Produce Uni-
formity in a Body,” ibid., 7 [2J, 82-85 (1924).
5. Discussion on “Sampling Fire Clays,” Bull. Amer. Ceram. Soc., 3 [6], 230 (1924).
6. Standards Report, “Standard Method for Sampling Ceramic Materials as De-
livered,” Jour. Amer. Cerum. Sac., l l [6J,442 (1928).
7. See also references, ibid.,p. 44 I .
OXFORD
MININGA N D MILLINGCOMPANY
WEST PARIS,MAINE

NOTES ON TUNNEL KILN CAR LUBRICATION*

BY J. T. ROBSON
ABSTRACT
A discussion of the properties required by oil for kiln-car lubrication is presented.
The results reported on questionnaires answered by forty plants operating tunnel kilns
are given, together with comments and conclusions therefrom. The residue, flash,
and burning points of ten different oils approved for kiln-car lubrication are listed with
directions for using oil as a lubricant on tunnel kiln cars.

I. Introduction
The consensus of opinion among oil producers and kiln-car users is that
the flash and burning points of the lubricating oil are of the most vital
importance. This may be true in a few rare cases where the temperature
of the car bearings is sufficiently high t o cause the oil t o flash or burn,
but even in such cases the heating-up is usually a gradual process so
that actually the oil undergoes a distillation and by the time the maximum
temperature is reached, there is not sufficient lubricant left to flash or
burn; there is only the carbonaceous residue. It is this residue which
has been found to be the cause of sticking or freezing of the roller bearings
and consequent hard running of the wheels.
I n general, the heavier the oil, the higher the flash and burning points,
but also the greater the percentage residue, and, while the oil may be
sufficiently heavy so t h a t i t does not all distill out while passing through
the kiln the first time, when it has distilled out and left its residue, the
amount of residue will be greater in proportion t o the number of times the
car has passed through the kiln than would be the residue from a lighter
oil with lower flash and burning points and lower residue. This is only a
general condition that applies to the ordinary rule : the heavier the oil
* Received February 23, 1932.